93 results about "Event graph" patented technology

The invention discloses an event atlas construction system and method based on multi-dimensional feature fusion and dependency syntax. The event graph construction method based on multi-dimensional feature fusion and dependency syntax is realized through joint learning of event extraction, event correction and alignment based on multi-dimensional feature fusion, relationship extraction based on enhanced structured events, causal relationship extraction based on dependency syntax and graph attention network and an event graph generation module. According to the event graph construction method and device, the event graph is constructed through the quintuple information of the enhanced structured events and the relations between the events in four dimensions, and the defects that in the priorart, event representation is simple and depends on an NLP tool, the event relation is single, and the influence of the relations between the events on event graph construction is not considered at the same time are overcome. According to the event atlas construction method provided by the invention, the relationships among the events in four dimensions can be randomly combined according to different downstream tasks, and the structural characteristics of the event atlas are learned to be associated with potential knowledge, so that downstream application is assisted.

A method for analyzing network incidents within a network and providing prioritized remediation recommendations is disclosed. The method includes: receiving network data and computing a plurality of network incidents from the network data, collecting network incidents related to a particular network issue over a time period and grouping the network incidents according to root-cause symptoms, generating a network incident graph by superimposing groups of network incidents over a network graph, analyzing the network incident graph to identify localized areas with systemic issues; and based on the analysis of the network incident graph, generating and displaying a list of remediation recommendations. Each remediation recommendation includes a systemic issue in the network, a remediation to resolve the issue, and a quantified expected benefit from implementing the remediation.

The invention relates to a domain event graph construction method and device fusing multiple types of facts and entity knowledge. The method comprises the following steps: performing cause relationship extraction and instance cause element extraction on a domain corpus to form an instance cause logic knowledge base; constructing an abstract concept knowledge base with hierarchy; performing entityword abstraction and predicate abstraction on instance events in the instance cause logic knowledge base by utilizing the abstract concept knowledge base to form an abstract cause graph; carrying outentity linking on instance events in the instance cause logic knowledge base by utilizing the entity knowledge graph and adopting an entity linking technology, and fusing event knowledge and entity knowledge to form a cause knowledge graph; and combining the abstract factorial map with the factorial knowledge map to form a domain event map. The advantages of static entity knowledge and action event knowledge can be integrated, the application range of knowledge questions and answers can be widened, and the domain event graph can be used as a common knowledge base to expand domain language resources.

The invention relates to an information classification extraction method and device based on a classifier, computer equipment and a storage medium. The information classification extraction method comprises the steps of obtaining to-be-classified information; performing word segmentation processing on the to-be-classified information to obtain a plurality of text blocks; inputting the plurality oftext blocks into a trained classifier, classifying the plurality of text blocks, and determining a target event category and a plurality of event elements of the to-be-classified information; extracting corresponding event element parameters according to the target event category and the event elements, and outputting the target event category, the plurality of event elements and the corresponding event element parameters; and generating event atlas data corresponding to the target event category according to the plurality of event elements and the corresponding event element parameters. By adopting the information classification extraction method, the event information can be accurately and effectively classified, and the key event information can be extracted.

The invention provides an event map construction method and device, and electronic equipment, and relates to the technical field of information mining, and the method comprises the steps of obtainingmultiple target event texts of a target event; performing meta-event extraction on the target event text to obtain a plurality of original meta-events; wherein the original meta event comprises a first event keyword; determining a plurality of target meta-events matched with the fact map from all the original meta-events based on the preset fact map and the first event keyword; wherein the cause map comprises events connected according to an event causal relationship and/or an event time sequence relationship; and updating the events in the cause map according to the target meta-event to obtain an event map corresponding to the target event text. According to the invention, the complexity of generating the event graph can be reduced, and the obtaining efficiency and event tracking capability of the event graph are improved.

The invention discloses a dam safety operation-oriented event atlas construction method. The method comprises the following steps of: (1) embedding enhanced semantic information by using word vectors; (2) introducing local attention and global attention to capture keywords and mine document information; (3) solving a sample imbalance problem by using Focalloss and extracting an event type; (4) adding event type codes, connecting the event type codes in series to form a new sentence code, and processing embedded vectors after series connection by using BiLSTM; (5) converting the attention network and the attention network according to the proportion, extracting effective features, and capturing corresponding event arguments; (6) filling an event missing document level argument by using an argument filling network; and (7) converting the attention network and the attention network labeling sequence by the weighted fusion graph, obtaining a causal relationship between events, and constructing an event graph.

Provided are systems and methods for generating an event map of events located near a user's location. The event map may include embedded media, such as audio, video, images, and the like for events located within a specific geographic area, such as geographic radius around a user's location, and a specific timeframe. Each event may be displayed at a position on the event map that corresponds to the location of the event, and the embedded media for the event may be displayed at an event's position. Additionally, the event map may include data for each event, such as date and time of the event and links to event-related resources.

The invention provides an event portrait-oriented text analysis method for a dialogue system. The event portrait-oriented text analysis method comprises the following steps of: 1, extracting an event relationship; 2, event extraction and structuring; 3, event similarity calculation and fusion; 4, event trigger word extraction and argument extraction. According to the method, the method for constructing the event graph is fully researched, the logic structure relation between events is described by adopting a knowledge graph structure, and meanwhile, in order to describe the events more clearly, the multi-dimensional attributes of the events are portrayed. The invention provides a system structure constructed by a affair graph. The structure comprises text corpus collection, event relation extraction template construction, event extraction and structuring, event similarity calculation and fusion, event trigger word extraction and argument extraction model construction, event relation pair construction and graph database storage. In addition, by utilizing the system structure, a sequential affair map, a causal affair map, a conditional affair map, a turning affair map and a concurrent affair map are constructed.

The invention relates to the technical field of data processing, is applied to the financial industry, and particularly relates to a data search matching method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a keyword to be searched and an event graph corresponding to the keyword; according to the relationship among the nodes in the event graph, calculating the distance between the node to which the keyword belongs in the event graph and other nodes in the event graph, and extracting the nodes of which the distances are smaller than a first preset threshold value to obtain a first expansion keyword; according to the relationship among the nodes in the target encyclopedia atlas, calculating the distance between the node to which thekeyword belongs in the target encyclopedia atlas and other nodes in the target encyclopedia atlas, and extracting the nodes of which the distances are smaller than a second preset threshold value to obtain a second expansion keyword; and combining the keyword, the first extended keyword and the second extended keyword to obtain a keyword set, and performing search based on the keyword set and a preset search and analysis engine to obtain a search result.

The invention relates to an abnormity detection method based on a log event graph and association relationship mining. The method comprises the following steps: collecting an original log of a system to obtain log events; segmenting the log events into different groups according to a set time span or a task number, wherein the log events in each group form a log event sequence according to the generated time; according to association relationship mining, mining system log events having association relationships with each abnormity, and removing log events irrelevant to the abnormities in the log event sequence; extracting a semantic vector of each log event as a feature vector of the log event; generating a bidirectional full-connection log event graph according to the log event sequence, updating a feature vector of each node by using a gating graph neural network, performing weighted summation on the updated feature vectors of all the nodes by using an attention network, calculating a global feature vector of the log event graph, finally performing classification detection through a full-connection network, and obtaining the normal or abnormal type of the system.

An automated technique for security monitoring leverages a labeled semi-directed temporal graph derived from system-generated events. The temporal graph is mined to derive process-centric subgraphs, with each subgraph consisting of events related to a process. The subgraphs are then processed to identify atomic operations shared by the processes, wherein an atomic operation comprises a sequence of system-generated events that provide an objective context of interest. The temporal graph is then reconstructed by substituting the identified atomic operations derived from the subgraphs for the edges in the original temporal graph, thereby generating a reconstructed temporal graph. Using graph embedding, the reconstructed graph is converted into a representation suitable for further machine learning, e.g., using a deep neural network. The network is then trained to learn the intention underlying the temporal graph. The approach operates to understand running behavior of programs, to classify them, and then enable detection of potential malicious behaviors.

The invention discloses a time sequence point-based network attack event prediction method, system and device and a medium, and the method comprises the steps: obtaining historical attack event data, generating a first event sequence according to the historical attack event data, and determining a first event feature and event occurrence time; constructing a first event atlas according to the first event feature, and processing the first event atlas through a graph embedding algorithm to obtain a first feature vector; inputting the first feature vector and the event occurrence time into a pre-constructed recurrent neural network for model training of a depth point process to obtain a trained time sequence point process model; and predicting the occurrence time and the event type of the next network attack event according to the time sequence point process model. According to the method, the event atlas is converted into the feature vectors through graph embedding, the feature vectors and the occurrence time are trained through the depth point process, the accuracy of network attack event prediction is improved, and the method can be widely applied to the technical field of network security.

The embodiment of the invention discloses an information recommendation method and device, and relates to the field of artificial intelligence. According to the specific implementation scheme, the method comprises the steps: enabling a server to use sample information to train a classification model in advance, employing the classification model for mining affiliation between events in an event database to obtain an event graph, and storing the affiliation, wherein nodes in the event graph are the events, and if edges exist between the events, it shows that the two events are affiliation; after a recommendation request sent by a user through the terminal equipment is received each time, determining a first event in which the user is interested according to historical browsing records and the like of the user, determining a second event which is a subordinate event with the first event on the basis of the first event and the event map, and sending target information containing the second event to the terminal equipment. By the adoption of the scheme, whether the two events are subordinate events or not is determined through the classification model trained through the characteristics of the events, the purpose of mining the subordinate relationship between the events is achieved, and then accurate recommendation is achieved.

The invention provides a movie script plot prediction method. The method comprises the following steps: segmenting a historical text into sentences; determining an event contained in the statement andan event feature of the event; establishing an event graph according to the time sequence of each event; wherein each node in the event graph corresponds to one event, and the feature vector of the node is determined through the event feature of the event corresponding to the node; and inputting the feature vectors of the nodes contained in the specified node path into the trained graph neural network, and outputting the prediction node of the event corresponding to the specified node path. By adding the event characteristics of the event, the event information is enriched, so that the accuracy of movie theatre plot prediction is improved.

The invention provides an information event mapping system and method based on natural language processing. The system comprises an information classification screening module, a part-of-speech tagging and grammar analysis module, a text syntax tree construction module, a subject-called-object analysis and extraction module, a multi-sentence entity anaphora resolution module and a knowledge graphdisplay module. By implementing the method and the system, analysts can be helped to know key information of the information more visually, so that the information reading work efficiency and the analysis accuracy of the analysts are improved.

The invention relates to a power grid accident information automatic filling and auxiliary decision-making method based on an event graph. The method comprises the following steps: firstly, performing natural language analysis on a text of a power grid accident record, performing information extraction, obtaining a relationship between accidents and a processing condition of the same type of accidents, combining an OMS, realizing automatic classification and input of accident information according to classification labels such as an accident occurrence area range and an accident type, then constructing an ontology by using an OWL language, inputting the extracted information, constructing an event ontology, communicating the events by taking the transition probability between the events as a connection, and establishing an event graph model. According to the invention, based on the theory and method of the event graph, on the basis of a power grid accident record text, rapid, accurate and comprehensive power grid fault information extraction and automatic filling are realized through a natural language analysis technology, an instantiated power grid accident record event graph is constructed, and various functions of information search, auxiliary decision making, accident analysis and the like can be provided for operators.

The technology disclosed includes a system to reduce clutter during graph presentation for security incident analysis. The system includes logic to score nodes potentially collapsed by equivalence, of indicated interest for security incident analysis, to prevent aggregation. The system includes logic to aggregate and hide equivalent nodes that have matching degrees, that are connected to matching nodes by matching edge types, and that have scores below a first selected threshold. The system does not collapse nodes that are interesting for security analysis and keeps them visible. The technology disclosed identifies chains of at least three nodes having degrees of 1 or 2, without branching from any node in the chain. The identified chains are collapsed into chain-collapsed single nodes. Two different cases of chains including whisker chains ending in a leaf node and chains connected at both ends to two other nodes are presented.

The invention discloses an event mode frequent subgraph mining and prediction method, relates to the technical field of knowledge engineering, and aims to reduce mining overhead and improve mining speed. According to the technical scheme, the method comprises the following steps: dividing nodes in a density graph abstract graph into clusters or super nodes in a density-based graph abstract stage,and sequentially selecting the nodes to construct a concise high-level graph; in the mode mining stage, carrying out frequent sub-graph mining on a large-scale event graph, mining frequent sub-graphsbased on the event mode, and finding frequently appearing sub-graphs in a graph set of the event graph; abstracting the input graph G based on a graph abstract algorithm, and performing frequent sub-graph mining and graph abstract-based mining preprocessing by taking an abstract result as input; and finally, mining and predicting an event mode from multiple sources and multiple perspectives by utilizing multi-source data, and performing candidate set filtering and frequent sub-graph output according to a minimum support degree min_sup defined by a user or other output standards.

The invention discloses a video description method and system based on a multi-level codec. The method comprises the steps: obtaining a video, carrying out the coding processing based on a multi-level encoder, and constructing a target graph and an event graph; and decoding the target graph and the event graph based on a multi-level decoder to obtain a sentence sequence and a word sequence, and completing a text description task and a sentence label prediction task based on multi-task learning. The system comprises an encoding module and a decoding module. Based on the multi-level encoder and the multi-level decoder, the relation of fine grit in the statement can be mined, and the description performance can be improved. The video description method and system based on the multi-level codec can be widely applied to the field of video processing.

Provided is a time-space dimension based event graph splicing method used for an internet public-sentiment system. The method utilizes a HBase column storage oriented non-relational database to storelarge amounts of internet public-sentiment event fragment information in a (space, time and key word) mode, quick event graph splicing according to time-space attributes is completed by proposing a time-space index structure and designing a corresponding query method. Compared with the other inventions, all fragment information is comprehensively organized according to time and space so as to completely show a whole event, accordingly a user can comprehensively and systematically know the time-space cause and effect development of the event, technical support is provided for development prediction and disposal decision making of internet public-sentiment events, and powerful assistance is provided for achieving social management functions.

The invention provides an event graph matching method and device, electronic equipment and a storage medium, and the matching method comprises the steps: obtaining a target event graph and a pluralityof to-be-matched graphs, carrying out the decomposition of the target event graph based on a preset dimension, obtaining a plurality of target event sub-graphs, and carrying out the decomposition ofeach to-be-matched graph based on the preset dimension, obtaining a plurality of to-be-matched sub-maps corresponding to each to-be-matched map, and respectively obtaining the similarity between eachtarget event sub-map and the to-be-matched sub-map under the same dimension in each to-be-matched map so as to determine the map similarity between the target event map and each to-be-matched map; anddetermining the to-be-matched atlas corresponding to the atlas similarity in a preset similar event similarity interval as a similar event atlas matched with the target event atlas. Therefore, the atlas similar to the event atlas can be accurately found, the calculation complexity is reduced, and the atlas comparison accuracy is improved.